نقش مقایسه اجتماعی در برداشت دانشجو از توانایی: نمای غنی از انگیزه علمی در دانش آموزان مدارس متوسط

Abstract This study addressed notions derived from a model by Tesser, Campbell, and Smith (1984) of self-concept and how it is influenced by social relationships. We were interested in whether the self-evaluation maintenance model (SEM) would allow us to investigate more directly the value component in expectancy–value models of achievement motivation. Using distinctions often made about different motivational orientations, we discovered a three-way interaction between level of mastery orientation (high or low), relevance of math (high or low), and target of rating (self or 7th grade friend) on students’ perception of ability in math. Thus, the SEM model was supported but only for students reporting a low mastery orientation. For them, those who reported math as highly relevant estimated their own ability as significantly higher than their friend’s, whereas those reporting math as less relevant showed no difference in estimates of ability between self or friend. For high mastery oriented students, no differences were found.

1. Introduction Parents and educators often express concerns about the real and apparent lack of interest and intrinsic motivation for academic endeavors that children demonstrate, particularly as they make the transition from elementary to middle school grades. Explanations of such a lack of interest in school learning run to several sources but often, the role of social influence and social comparison comes into the picture (Berndt & Keefe, 1996). In this view, schooling is a context in which children are easily made aware of how they and others perform, of how important performance is as opposed to learning, and these concerns with performance have a negative impact on intrinsic motivation. One potential source of explanation of this social comparison phenomenon is a model developed in the social psychology literature that relates social comparison to academic performance. Tesser and Smith (1980) proposed a model of how individuals maintain their self-view by evaluating the importance they place on their own performance and ability in particular domains relative to their peers’ performance and ability. Thus, in our study we were interested in whether students’ academic motivation orientation would mitigate the effects of processes by which this self-evaluation takes place. Specifically, we speculated that the claims of the SEM model may not apply to students who are high in mastery/learning orientation because for them, the accomplishments of others are not relevant to what propels them to achieve their academic goals. The claims of the SEM model that relate to how one perceives how others are performing on a task would seem to apply to students whose academic motivation is based on comparisons of self to others, what is at the heart of different kinds of performance orientations. 1.1. The self-evaluation maintenance (SEM) model In order to summarize how one’s perceptions of peers’ and strangers’ performance in a particular domain influence one’s self-evaluation, Tesser and Smith (1980) proposed a model, named the self-evaluation maintenance (SEM) model, that has proved generative in the social psychological literature but has not been associated with the prolific educational psychology literature on academic achievement motivation. The model describes a dilemma faced by friends when they consider their performance in a particular domain, the relevance of that domain to their self-concept (by which Tesser and colleagues meant how individuals view themselves), and their closeness to each other. For example, a student who perceives that she is performing better than others in a particular academic domain that is relevant to her will likely pursue friendships only with those whom she perceives as less able than she is in that domain. Or, from a different perspective, a student may decide that a close friend is much better than he is in an academic domain and as a result, decrease the relative importance of that domain and his motivation to engage in academic activities. The SEM model has been investigated in conjunction with several other psychological constructs, including physiological arousal (Achee, Tesser, & Pilkington, 1994), jealousy (DeSteno & Slovey, 1996), narcissism (Morf & Rhodewalt, 1993), and social identity. Tesser and colleagues have applied the SEM model to the cognitive processes of students in an educational context. Students frequently experience situations in which one or more aspects of school are highly relevant to their sense of self and typically experience psychological closeness with classmates or friends. In two studies, Tesser and colleagues used self-report measures to examine the interrelations among performance in school, friendship choices, and school-related activities that helped define a student’s sense of self, and found support for the SEM model. In the first study, Tesser and Smith (1980) assigned college students to two conditions. The target student was either paired with a friend or with a stranger and was asked to give the partner clues for solving either a high relevance task (important academic skills) or low relevance task (game). Tesser and Smith discovered that student participants gave less helpful clues under high relevance than under low relevance conditions and they gave less helpful clues to strangers than to friends in the low relevance condition. In the second study, Tesser et al. (1984) asked fifth and sixth graders to nominate their friends in school. The target students’ nominations were largely friends whose perceived performance was better than their own on irrelevant school activities and inferior to their own on relevant school activities. In actuality, the target students and their friends had very similar performance levels in the relevant school activity. 1.2. Limitations of the SEM model The SEM model clearly suggests that the process of social comparison is necessary for maintaining self-evaluation. Tesser and Campbell (1982) outlined the comparison process as having two possible outcomes. The first possibility is that self-evaluation can be enhanced by reflection processes when a close other performs well. The second possibility is that self-evaluation can be decreased as a result of comparison if a close other’s success makes our own performance look bad in comparison. While social comparison is a specific component of the SEM model, social influence, which may explain the larger process of how friends influence each other, is not. According to Berndt and Keefe (1996), self-enhancement is a motive related to social comparison that falls under a distinct pathway of peer influence. The pathway is indicated by the influence of attitudes, behavior, and other characteristics of friends. Other motives that account for peer influence are the need for approval, identification with friends, the need to be correct, as well as the need for self-enhancement. The first motive, students’ need for social approval, is marked by behavior that will meet friends’ expectations or make a positive impression on friends. The second motive, identification with friends, is the desire of students to think and behave like their friends. The third motive, the need to be correct, is related to a person’s desire to hold correct beliefs and make reasonable decisions. Finally, self-enhancement is indicated by a student’s need to judge his or her own competence by comparing his or her performance with that of a close friend. Berndt and Keefe acknowledged that Tesser’s SEM model suggests a form of social comparison that meets a student’s need for self-enhancement, but argued that students are not completely free to define domains of achievements as unimportant. For example, teachers and parents may not agree with a student that getting a C in math is unimportant and direct comparisons with peers are often made in the classroom whether the student likes it or not. Therefore, there may be other sources of peer influence beside the process of social comparison that are motivating students to make certain decisions about their friendship choices and/or their performance in school. While the SEM model appears to have its place in explaining interesting classroom phenomena, there have been theoretical and empirical challenges indicating the limitations of the construct. For instance, Manion (1992) found that the SEM model was only applicable in competitive, not cooperative, academic situations. Also, Corno and Mandinach (1983) critiqued the SEM model on the grounds that the causal paths between social comparisons and motivated behaviors are not presented in clear directions, something they felt their self-regulation model captured more effectively. To our knowledge, the SEM model has so far not been tested empirically along with any of the numerous constructs from the academic achievement motivation literature, an oversight that is curious given its clear connection to issues of motivation. In particular, we were interested in examining how well the predictions that derive from the SEM model would apply to individuals with different motivational orientations. 1.3. Academic achievement motivation and SEM Academic achievement motivation has historically been defined as goal-directed behavior (Graham & Weiner, 1996). To that definition, recent literature has added a broader, more conceptual framework for organizing both cognitive and affective components of motivation orientation (Ames, 1992; Elliot & Harackiewicz, 1996; Pintrich & Schunk, 1996). In this view, achievement goals in general define patterns of motivation that represent different ways of approaching, engaging in, and responding to achievement-related activities. For instance, Ames (1992) characterized academic goals as mastery or performance orientations, Nolen (1988) and Nicholls (1984) as task or ego oriented goals, while Dweck (1986) used the performance or learning orientation contrast. These constructs of motivation are derived historically from attributional models of motivation, and as Pintrich and Schunk state, “attribution theory and research on attributional processes… have a direct link to individuals’ expectancy beliefs” (p. 107). Originally proposed by Atkinson (1964), the expectancy–value theory assumes that for all individuals, achievement situations arouse both positive and negative motives, with the fear of failure and the desire for success competing with each other. Expectancies result from the cognitive appraisals of a situation leading the individual to make a probabilistic judgment of the likelihood for success in a particular situation. Value refers to the relative attractiveness of success in a particular situation, and thus, one tends to avoid tasks that are negatively valued and approach tasks that are positively valued (Wigfield & Eccles, 1992). The expectancy side of the theory has been examined much more thoroughly and closely, while the value component has been relatively neglected in both theoretical and empirical work on achievement motivation (Brophy, 1999; Wigfield & Eccles, 1992). Although rare, there have been some studies that have incorporated the value side of the expectancy–value equation. For instance, in a model established to explain the relationship of achievement-related values, Eccles and her colleagues (1983) explained that students’ expectancies and values have the most direct effects on their performance and choice of achievement tasks. Expectancies and values themselves are influenced by children’s goals and task-specific beliefs, beliefs that are perceptions of competence and perceptions of the difficulty of different tasks. This includes the concept of attainment value, defined as the importance that an individual places on a task, and can be further broken down into two subcomponents: the importance of a given activity (absolute attainment value) and the importance of it relative to other activities (relative attainment value) (Battle, 1966). Wigfield and Eccles have expanded upon Battle’s original conception of attainment value to include the relevance of engaging in a task to confirm or disconfirm salient aspects of one’s self-schema. That is, they argued that tasks provide the opportunity to demonstrate aspects of one’s actual or ideal self-schema. Therefore, tasks will have higher attainment value to the extent that they allow the individual to confirm salient aspects of these self-schemata. This component of the expectancy–value model seems most similar to Tesser’s SEM model, in which students evaluate their performance in a particular domain and the relevance of that domain to their self-concept (Tesser et al., 1984). From a developmental standpoint, it seems appropriate to examine the SEM model in the context of middle school students’ academic achievement. At this point in a child’s development, friends become a very important source of information about what is to be valued in school and how to achieve that which is important (Berndt & Keefe, 1996; Berndt, Lyachak, & Park, 1990). Also, it is known that students’ valuing of academic subjects typically declines across the transition from elementary school to junior high or middle school. For example, according to Eccles et al. (1989), students’ self-concept of ability and importance ratings for math show a linear decline in the transition from elementary school to junior high or middle school while students’ perceived interest, or liking in math, remain the same. In a recent study conducted by Anderman et al. (2001), students’ self-concept of ability in math was related positively to increases in achievement values. Interestingly enough, perceived ability, perceived importance, and liking of academic domains are all elements of Tesser’s SEM model, all aspects that seem to address the value students place on academics. We agree with Wigfield and Eccles (1994) that it is important to assess specific aspects of children’s subjective task values about different activities. Therefore, we examined Tesser’s SEM model in 7th grade math classes in the context of Dweck’s model of motivation orientation. (Note that Dweck’s distinction has been elaborated on, revised, and re-labeled but the basic construct of different goal orientations remains throughout these new conceptions; see Ames, 1992; Maehr & Midgley, 1991; Nicholls, 1984 and Nicholls, 1990). The distinction between performance and mastery orientations seemed promising as a way to understand the limits of the SEM model. According to Dweck’s original conception, children with learning goals typically have a mastery oriented approach to academic tasks, and are eager to seek challenge regardless of their perceived ability for the task. Children who have performance goals seek approbation for their successes (what, in modern formulations of goal theory is called performance-approach) and work to avoid public failure (what is now called performance-avoid, Elliot & Harackiewicz, 1996). Thus, we conducted a study by asking all the 7th grade students in a local middle school to fill out questionnaires that would allow us to determine their perceptions of theirs and their friend’s ability in math, their motivational orientation in math, and their perception of the relevance of math for themselves and their friends. The research questions were as follows: 1. How does a student’s self-rating of ability in math differ from his or her rating of a close friend’s ability, when considering the relevance of math to the student (high versus low) and the student’s level of mastery orientation (high versus low)? 2. Are the same variables as tested in Research Question 1 of motivational orientation and relevance of math important in distinguishing the actual math achievement of a student and his or her friend?

3. Results Confirmatory factor analysis was performed on the revised version of the Goals Inventory Instrument to see if the modified instrument had the same measurement characteristics reported for the original instrument by Roedel et al. (1984). Specifically, SAS’ PROC CALIS was used to see if the orthogonal structure specified by Roedel et al. fit the data obtained in this study. Because we obtained a goodness of fit index (GFI) of only 0.82, we concluded that the modified instrument did not possess the previously reported measurement characteristics when applied to our population of 7th grade middle school students. Additionally, the 90% confidence interval for the RMSEA estimate did not contain the accepted criterion, 0.05, for “close fit.” Given this relatively poor fit, we decided to use exploratory factor analysis to ascertain the measurement characteristics of the instrument as it applied to our population. The exploratory factor analysis was conducted using the principal factors method within SAS’ PROC FACTOR. Two factors were retained using the eigenvalue greater than one criterion. This two-factor solution reproduced the sample correlation matrix reasonably well in that the root mean square off-diagonal residual was equal to 0.06. Also, 89% of the common variance was explained by the two factors. After performing a non-orthogonal rotation, the two factors that emerged from this rotation could clearly be labeled as learning and performance orientation with a correlation index of −0.44 (see Table 2 for factor loadings). Table 2. Factor loadings for mastery and performance orientation from the adjusted goals inventory scale Item Factor loading Factor 1: Mastery orientation 1. I enjoy challenging school assignments in Math .45 3. I work very hard even when I am frustrated by a task in Math .56 6. I try even harder after I fail at something in Math .63 7. I adapt well to challenging circumstances in Math .46 9. I work hard even when I don’t like Math class .66 10. I am very determined to reach my goals in Math .77 11. Personal understanding of Math is very important to me .72 12. I work very hard to improve myself in Math .83 16. I am naturally motivated to learn in Math .62 17. I prefer challenging tasks in Math even I don’t do well at them .53 22. I feel most satisfied when I work hard to achieve something in Math .68 Factor 2: Performance orientation 2. It is important for me to get better grades than my classmates in Math .58 13. I like others to think I know a lot about Math .54 14. It bothers me the whole day when I make a big mistake in Math .42 16. I feel angry when I do not do as well as others in Math .32 Note. Students were asked to respond to each item choosing from five-point Likert scales with the following response choices: 1, not at all like me; 3, does not apply to me; 5, very much like me. Table options Research Question 1: What is the interaction of mastery orientation (high versus low), relevance of math (high versus low), and focus of rated ability (self versus friend) on perception of ability? Two repeated measures ANCOVAs were used with perception of ability as the dependent measure (see Table 3). The design was a 2×2×2 analysis with relevance (high or low relevance of math to self) and mastery (high or low mastery orientation) as between-subjects variables, and focus of rating (self versus friend) as a within subject variable. In one analysis, the focus of rating compared self with close classmate in math. In the second, the focus of rating compared self with 7th grade friend. Students’ liking of math, which correlated moderately with perception of ability (r2=.42, p<.01), was added as a covariate. Table 3. Descriptive data for target students’ report of perceived ability in math and students’ actual math grades Level of math relevance and mastery for target student n Target’s perception of abilitya Gradesb Self Class-friend 7th grade friend Self Class-friend 7th grade friend Non-friend High relevance/high mastery 111 M 3.92 3.88 3.91 88.36 88.32 88.70 86.64 SD 0.77 0.90 0.87 6.45 6.64 6.11 6.78 High relevance/low mastery 11 M 4.00 4.60 3.67 88.64 86.14 83.70 85.40 SD 0.82 0.89 2.31 5.52 9.87 7.25 5.32 Low relevance/high mastery 43 M 3.33 3.71 3.67 85.77 87.50 87.43 87.55 SD 0.80 0.86 0.58 6.54 6.60 6.47 6.28 Low relevance/low mastery 37 M 3.13 3.67 3.50 84.76 86.77 86.74 88.94 SD 1.10 0.91 0.52 6.63 8.01 6.60 6.57 a Means range on a rating scale of 1–5, 5 being the highest. b Means range from 0 to 100. Table options We were able to come to similar conclusions, as indicated in Tesser, Campbell, and Smith’s findings (1984), for both a close classmate and for a 7th grade friend with a significant interaction effect between relevance and target (self versus friend) (F(1,139)=9.93, MSerror=.53, η2=.067, p<.05; F(1,146)=14.33, η2=.089, p<.01). Pairwise comparisons using least significant difference (LSD) estimates revealed that students for whom math was not relevant rated the ability of their close classmate and of their 7th grade friend as significantly higher than their own (p<.05). By contrast, students for whom math was very relevant rated themselves as significantly higher in ability than their close friends (p<.05). We found a significant three-way interaction with mastery (high/low) for the 7th grade friend analysis (F(1,146)=4.26, MSerror=.85, η2=.028, p<.05). Pairwise comparisons revealed that for the 7th grade friend analysis, students with low mastery showed the same Tesser effect as we found in the two-way interaction but students with high mastery did not (p<.001). The same three-way interaction for the analysis using the close classmate in math as a comparison was not significant. Research Question 2: What is the interaction of level of mastery orientation and relevance on self versus friend’s actual math grades? In contrast to Question 1, in which students’ perceptions of their math ability was the dependent variable, Question 2 tested the effect of the same independent variables on actual math grades with three repeated measures ANOVAs. The design was a 2×2×2 analysis, with relevance (high or low relevance of math to self) and mastery (high or low mastery orientation) as between-subjects variables and target versus friend/non-friend as a within subject variable. In one analysis, the within subject variable compared self and the close classmate’s grades. In the second analysis, the comparison was between self and the distant classmate. In the third, the comparison was between self and the 7th grade friend. We were able to match Tesser’s findings with the 7th grade friend and the non-friend analyses with a significant interaction between relevance and target (F(1,119)=4.29, MSerror=35.1, η2=.035, p<.05; F(1,115)=5.27, MSerror=39.0, η2=.044, p<.05), but not for the close classmate analysis. When considering mastery orientation, there was no evidence of a three-way interaction for target × relevance × mastery for either the 7th grade friend or the non-friend analysis. However, it is interesting to note that a significant two-way interaction effect was found between target × mastery for the target versus 7th grade friend analysis (F(1,119)=4.85, MSerror=35.1, η2=.039, p<.05). Pairwise comparisons revealed that the source of the interaction was the difference in grades between self and 7th grade friend in the low mastery group (p<.05). In other words, students with low mastery orientation tended to have 7th grade friends with significantly lower math grades.